This meta-analysis is the first to quantify the dose-response relation between physical activity and CHD risk with regard to both physical activity amount and magnitude of lower CHD risk. We found that individuals who met the basic US physical activity guideline for health9
had a 14% lower risk of CHD, compared to those with no leisure-time physical activity. Those meeting the advanced guideline had a 20% lower risk of CHD. At higher levels of physical activity, modest increments of risk reduction were observed. We also noted lower relative risks among persons who were physically active below the basic guideline, supporting the guideline’s assertion that some physical activity is better than none.
Interestingly, we observed a significant interaction by gender, such that the association of physical activity and CHD risk was stronger in women than in men. We were unable to assess whether the association differed by race or age, because of insufficient variation among studies. Geographic region of origin did not influence the association.
It is unclear why we observed a significant interaction by gender. Possible explanations include biologic differences, methodologic considerations, or some combination of both. Previous evidence does not support more favorable effects of habitual physical activity on CHD risk factors (including blood pressure, lipid levels, vascular indicators, cardio-respiratory fitness, and metabolic syndrome) among women compared to men.8
The type or intensity of physical activity contributing to total LTPA energy expenditure may differ between men and women (e.g., men favor vigorous activities while women are more likely to engage in moderate activities).17, 43
However, this does not explain the stronger effects in women, as there are limited data suggesting that vigorous-intensity physical activity may be associated with additional cardiovascular benefits, beyond its contribution to energy expenditure.47
Methodological issues may explain a portion of the difference. For instance, women have lower CHD rates1
; thus, the presence of imprecisely measured or unmeasured plausible confounders (such as smoking habit and diet) may have a smaller effect in women than men.
There may be gender differences in the reporting of physical activities. However, it is unlikely that such misclassification would be greater among men than women since vigorous-intensity activities (in which men are more likely to engage) tend to be better reported than activities of lesser intensity.48
Of the studies included, longer duration of follow-up was more likely in studies of men, leading to greater potential for misclassification of energy expenditure. However, analyzing a subset of studies with comparable follow-up in men and women did not change our main results.
The primary strength of this study was the quantification of physical activity amount in analyses, enabling assessments of the risk associated with specific quantitative levels of LTPA. We chose to quantify physical activity in units of kilocalories per week (and accounting for the different average weights of men and women) as they were more frequently reported in studies, and are a more easily understood unit. We also assessed potential effect modification by numerous variables, and reported gender specific results.
Although the selection of studies that included quantitative estimates of physical activity allowed for this more quantitative approach, it also limited the number of studies that could be included. In a secondary analysis we included several additional studies for which we were able to crudely estimate quantitative levels of LTPA; findings were similar to the main analyses. We also examined the potential influence of single studies, and found that no one study changed results.
This study was limited by inclusion of only English language studies, possibly resulting in bias since statistically significant results may be more likely to be published in English. However, it is unclear whether inclusion of only English language papers does cause bias.49,50
By designating meta-analytic methods a priori
, we aimed to minimize any potential investigator bias due to preconceptions. However, it is possible that the a priori
designations, as well as subsequent interpretations, were subject to personal biases. Because this is a meta-analysis of observational studies, the potential for residual confounding and bias cannot be addressed through pooling. A primary source of potential residual confounding is likely to stem from confounding variables which were either unmeasured or insufficiently measured in the individual studies themselves. For instance, dietary intake was rarely assessed in the studies reviewed. In all studies included, physical activity was assessed by self-report; some misclassification of activity levels is probable and quantitative characterizations should therefore be considered approximate in nature.
We were only able to conduct our primary analysis on LTPA on 9 of 26 of potential studies. As result, there were insufficient data to assess potential interaction by several important factors (e.g., baseline age and race). Among women alone, it appeared that there was a marked and sudden decline in risk at five times the minimally recommended level of physical activity (). However, this data point was based on only 2 studies.
We contacted the authors of the remaining 17 studies to request unpublished quantitative physical activity data; however, little additional usable information was obtained, as many of these studies used qualitative categories to assess physical activity. The inclusion of only the 9 studies for quantitative analyses was unlikely to have biased results, since these 9 studies appeared representative of the broader group of 26 eligible studies. In initial analyses comparing “high” versus “low” physical activity, which included all 26 studies, findings were similar to those including only the 9 studies. Further, in comparing our findings with previous reviews, which quantified only the magnitude of lower relative risks but not the amount of physical activity required, the results are comparable. Our comparison of “high” versus “low” physical activity yielded a relative risk of 0.75 for CHD, similar in magnitude to several past reviews.2, 3, 7, 8
In conclusion, the present study provides quantitative data supporting the 2008 Physical Activity Guidelines for Americans which recommends the equivalent of 150 min/week of moderate-intensity physical activity for health, and 300 min/week for additional health benefits, as well as encouraging any amount of activity for those unable to meet the minimum. Future studies that quantitatively assess the dose-response relation between LTPA, as well as other types and features of physical activity, and CHD risk will help clarify the upper end of the dose-response curve and enable additional quantitative evaluations in future reviews, such as exploring potential differences by age and race. Additionally, individual participant meta-analyses conducted via collaboration among research groups, though resource intensive, can make use of existing studies to further clarify dose-response relationships.51
Physical activity clearly has been shown to decrease the risk of developing coronary heart disease (CHD). However, the dose-response relation (How much activity is needed? What level of risk reduction is associated with specified levels of activity? Does the risk continue to decrease at higher levels of activity?) is less clear. This is the first meta-analysis of epidemiologic studies to quantify the dose-response relation, examining both the specific amounts of physical activity and associated risk reductions for CHD (previous meta-analyses have quantified only risk reductions, but not the specific doses of activity required). We found that individuals who engaged in the equivalent of 150 min/week of moderate-intensity leisure-time physical activity (corresponding to the minimum amount recommended by the 2008 US federal guidelines) had a 14% lower CHD risk, compared with those reporting no LTPA. Those engaging in the equivalent of 300 min/week of moderate-intensity leisure-time activity had a 20% lower risk. At higher levels of physical activity, relative risks were modestly lower; for example, at five times the minimum recommended, there was a 25% lower risk. Persons who were physically active at levels lower than the minimum amount recommended also had a significantly lower risk of CHD. These findings provide quantitative data that support the 2008 US physical activity guidelines. They indicate that the “biggest bang for the buck” for CHD risk reduction occurs at the lower end of the activity spectrum: very modest, achievable levels of physical activity.